Unraveling the chromophoric disorder ofpoly(3-hexylthiophene)Alexander Thiessena, Jan Vogelsangb, Takuji Adachib,c,d, Florian Steinerb, David Vanden Boutc,d,and John M. Luptona,b,1

aDepartment of Physics and Astronomy, The University of Utah, Salt Lake City, UT 84112; bInstitut für Experimentelle und Angewandte Physik, UniversitätRegensburg, D-93040 Regensburg, Germany; and cDepartment of Chemistry and Biochemistry and dCenter for Nano and Molecular Science and Technology,The University of Texas at Austin, Austin, TX 78712-0165

Edited by Peter J. Rossky, The University of Texas at Austin, Austin, TX, and approved August 2, 2013 (received for review April 24, 2013)

The spectral breadth of conjugated polymers gives these materialsa clear advantage over other molecular compounds for organicphotovoltaic applications and is a key factor in recent efficienciestopping 10%. However, why do excitonic transitions, which areinherently narrow, lead to absorption over such a broad range ofwavelengths in the first place? Using single-molecule spectros-copy, we address this fundamental question in a model material,poly(3-hexylthiophene). Narrow zero-phonon lines from singlechromophores are found to scatter over 200 nm, an unprecedentedinhomogeneous broadening that maps the ensemble. The giant redshift between solution and bulk films arises from energy transfer tothe lowest-energy chromophores in collapsed polymer chains thatadopt a highly ordered morphology. We propose that the extremeenergetic disorder of chromophores is structural in origin. Thisstructural disorder on the single-chromophore level may actu-ally enable the high degree of polymer chain ordering found inbulk films: both structural order and disorder are crucial to materialsphysics in devices.

structure–property relations | conformational disorder | photophysics |organic solar cells

Despite half a century of research into organic photovoltaics(1), the promise of versatile paint-on solar-cell modules

based on conjugated polymers has prompted a present flurry ofactivity in the field (2–5). A particular appeal of such excitonicsolar cells is that very little material is needed to efficiently ab-sorb light. This is due to the fact that the oscillator strength ofprimary photoexcitations, electron–hole pairs with binding en-ergies far exceeding kT, is focused in the excitonic transition. Theobvious downside is that this concentration also means excitonictransitions are inherently narrow and usually offer only mediocrespectral overlap with the broad solar spectrum. How then canexcitonic solar cells be designed with appropriate spectralbreadth? Merely introducing energetic disorder in the underlyingexcitonic material should lead to low-energy traps, impedingcharge harvesting.Although the optical and electronic properties of conjugated

polymers are not perfectly suited to photovoltaics, their ab-sorption spectra are surprisingly broad despite the excitonicnature of the transitions. Moreover, the diversity in functionalcharacteristics revealed by varying processing conditions hasfueled the quest to formulate robust structure–property rela-tionships between the electronic and optical properties and theunderlying polymer structure (6–12). Polythiophene derivativeshave evolved into one of the chosen workhorse materials forsolar-cell research (13, 14). Early spectroscopic studies estab-lished extraordinary solvatochromic and thermochromic char-acteristics, which were attributed to large conformational changesin response to the immediate environment (15–17). It is littlesurprise then that such diversity in device characteristics existswhen using these materials (4). Poly(3-hexylthiophene) (P3HT)(structure shown in Fig. 1A) and related compounds were thefirst polymeric materials to exhibit signatures of 2D electronic

delocalization (18) due to a high level of interchain orderingmost clearly visualized in X-ray diffractometry (19). Despite thisordering, the optical absorption remains extremely broad, ren-dering the material favorable for photovoltaics. The strikingdependence of P3HT ensemble optical properties on processingconditions has led to proposals that substantial dimerizing in-terchain interactions (20–23) could be involved. However, thesignatures of excitonic dimerization, or H-aggregation, are notalways straightforward to resolve conclusively (24). H-aggrega-tion should lead to a reduction in radiative rate (25). Typically,upon aggregation of P3HT, a strong decrease in fluorescencequantum yield is observed, but this is accompanied by an in-crease in fluorescence rate (26) because nonradiative decay ratesare also enhanced. It is not always trivial to distinguish an in-crease in nonradiative decay from deceleration of radiative de-cay, complicating a precise determination of transition oscillatorstrengths. For this reason, most of the focus to date has been oninterpreting spectral shape and vibronic coupling (23). Althoughthe origin of the magnitude of spectral shift between isolated andbulk-packed chains has been simply assigned to a “crystal shift”(23), a high degree of interchain ordering should also implysubstantial planarization of the polymer with the associated in-crease in conjugation length (27)—which in turn decreases thestrength of dipolar coupling that could lead to dimerization asthe excited state becomes more delocalized (21).Conclusively discriminating interchain from intrachain order is

only possible by resorting to subensemble techniques such assingle-molecule spectroscopy (28–35). Here, we demonstrate thefeasibility of reconciling fundamental spectroscopy of P3HT, shown

Significance

Ideal photovoltaic cells would be black, absorbing all of theSun’s radiation, whereas Nature’s machinery for solar energyharvesting—photosynthesis—looks green. Organic semicon-ductor devices, based on molecular building blocks, lie con-ceptionally between the extremes of inorganic and photo-synthetic light harvesting. How can organic solar cells appearalmost black if they are based on molecular units? Using single-molecule spectroscopy, we identify the fundamental electronicbuilding blocks of organic solar cells and reveal that discretemolecule-like transitions scatter over the entire visible spec-trum. The fundamental molecular unit is narrowband, butdisorder induces a continuum reminiscent of that characteriz-ing highly ordered inorganic crystals.

Author contributions: J.V. and J.M.L. designed research; A.T., J.V., T.A., and F.S. performedresearch; J.V. and F.S. analyzed data; and A.T., J.V., T.A., D.V.B., and J.M.L. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. E-mail: john.lupton@ur.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1307760110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1307760110 PNAS Early Edition | 1 of 7

CHEM

ISTR

YPN

ASPL

US

in this study to be the most heterogeneous polymeric materialstudied to date on the single-chromophore level, with the exci-tonic single-chromophore picture (36–38) established for themost ordered polymers such as polyfluorenes (39) (PF), ladder-typepoly(para-phenylenes) (40, 41) (LPPP), polydiacetylene (42,43) (PDA), and poly(phenylene-ethynylene) (28) (PPE). On thelevel of single chromophores, P3HT behaves like other prominentmaterials, such as poly(phenylene-vinylenes) (41, 44) (PPVs), withthe exception of exhibiting giant variability in transition wave-lengths spanning almost 200 nm. The dominant photophysicsof the ensemble—the striking red shift between solution and filmphotoluminescence (PL)—is controlled by energy transfer to low-energy intrachain chromophores, in absence of the previouslyproposed “crystal shift” (23).

ResultsComparison of Ensemble and Single-Chromophore Spectroscopy.Nominally different conjugated polymer materials exhibit very

similar spectral characteristics on the single-chromophore level(28). Because long polymer chains can contain many chromo-phores (40), it is not always straightforward to ensure thata single chromophore is identified within the polymer. Com-parison with oligomeric model compounds, use of narrow-bandexcitation, and cryogenic cooling of the sample have, however,enabled a reproducible framework for the identification of singlechromophores (40). We begin by comparing the ensemble PLspectra of P3HT in dilute toluene solution (10−4 mM) and ina drop-cast bulk film in Fig. 1A. Going from solution to the solidphase shifts the spectrum by over 100 nm (0.4 eV) to the red andmodifies the spectral form. The bulk spectrum has been inter-preted to show a shift of oscillator strength from the 0–0 to the0–1 transition, assigned to H-aggregation (22), because bulk-phase P3HT exhibits a high level of structural ordering in elec-tron microscopy (5), scanning-probe microscopy (45), and otherelastic scattering techniques (11, 46, 47). The large spectral shiftbetween solution and bulk has been assigned to the occurrenceof a crystal shift in the H-aggregated chromophores embedded inan ordered environment of many other conjugated polymerchains (23). Such a large spectral shift between solution and filmis not observed in other common conjugated polymers. Six rep-resentative single-chromophore spectra, recorded at 4 K fromisolated chains directly deposited from toluene onto a SiO2 sur-face, are superimposed on the ensemble spectrum in gray. Thesingle-chromophore spectra all have the same shape, a compar-atively narrow asymmetric zero-phonon line with a low-energyacoustic phonon wing and a distinct vibronic band offset by180 meV. The reduction of disorder broadening on the single-chromophore level facilitates determination of vibronic frequen-cies compared with the ensemble. The observed vibronics can beassigned to the ring C–C stretch (171 meV) and the symmetricC=C stretch modes (179 meV) (48). The single-chromophorespectra, obtained under excitation at 458 nm, span a spectralrange of 195 nm (0.8 eV). In the red-most spectrum, excitationand emission are separated by 0.9 eV, and yet the same universal(41), narrow single-chromophore spectral shape is observed.To highlight the universality of chromophores, we plot 115

spectra in Fig. 1B, sorted by 0–0 peak energy (see Fig. S1 foralternative forms of presenting the data). Including the vibronicprogression, narrow spectral lines are found between 485 and775 nm. The general spectral characteristics, in particular theenergy of the dominant vibronic plotted beneath, remain un-changed regardless of peak energy. The distribution of peakpositions is plotted in a histogram in Fig. S2 and appears to betrimodal (see the discussion in SI Text for possible origins ofthis distribution).Interestingly, the 0–0 peak narrows slightly with decreasing

energy, as plotted in Fig. 1C. Such spectral narrowing is seen inPF (28) and PDA (49), where lower-energy transitions corre-spond to improved chain ordering (the formation of the β-phasein PF). The opposite is seen in PPV (50), where chain bendinginduces a red shift and an increase in spectral jitter. Over theentire range of chromophore energies, the ratio between 0–0 and0–1 PL peak intensities scatters substantially but does not varysystematically with transition energy (Fig. 1D). Strong variationsin the PL intensity of the vibronic progression between singlechromophores have been reported previously, even in nominallyrigid materials such as LPPP (51). Such variations arise due tothe slight distribution in ground-state molecular conformations,leading to differing structural relaxation energies and theresulting changes in the Franck–Condon progression, which be-come visible precisely because electronic and vibronic transitionsare so narrow at low temperature. Examples of the interchromo-phoric scatter in 0–0/0–1 peak ratio are given in Fig. S3.The single-chromophore PL spectra are identical in shape to

that of different conjugated polymers (28) such as LPPP, PPV,PDA, PPE, and PF as illustrated in Fig. S1, and exhibit the

0

1

500 600 700 800 900

PL

Inte

nsity

Wavelength (nm)

1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6

150

180

EV

ib(m

eV)

0-0 Peak Position (eV)

1.6

1.8

2.0

2.2

2.4

2.6

Ene

rgy

(eV

)

1.8 2.0 2.2 2.4 2.60

10

20

30

40

50

0-0

Pea

kW

idth

(meV

)

0-0 Peak Position (eV)1.8 2.0 2.2 2.4 2.6

0.0

0.5

1.0

1.5 0-1/0-0P

ea kR

at i o

SS

n

P3HT

0-1

0-0

A

B

300 K4 K

4 K

in toluene bulk filmsinglemolecules

C D

Fig. 1. Unraveling the spectral heterogeneity of P3HT using low-tempera-ture single-molecule spectroscopy. (A) Comparison of P3HT PL spectra indilute toluene solution (green) and in a drop-cast bulk film (red) at roomtemperature. Six representative low-temperature (4 K) single-moleculespectra (gray), consisting of a zero-phonon line and a vibronic progression,are shown as examples, spanning the spectral region from dilute solution tobulk film. (B) Normalized PL spectra of 115 single molecules, sorted by thepeak energy of the 0–0 transition. The same 0–1 vibronic transition, shiftedby 180 meV from the main peak, is seen for all spectra (Lower). (C) Corre-lation of 0–0 peak width with transition energy. The green circles representaverages over eight molecules, and the red line is a guide to the eye. (D)The intensity of the vibronic peak does not depend on chromophore tran-sition energy.

2 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.1307760110 Thiessen et al.

common blinking and spectral diffusion (spectral jitter; Fig. S4).This similarity in spectral form strongly suggests that the narrowlines observed over such a broad spectral range arise from iso-lated chromophores (38). Because narrow single-chromophorezero-phonon lines are observed as far to the red as 680 nm, weconclude that the high-energy side of the bulk spectrum (Fig. 1A,red curve) is appropriately described by the presence of isolatedrather than aggregated chromophores: the red shift of over 100 nmbetween solution and film can be mostly attributed to energytransfer from high-energy to low-energy chromophores. Thisconclusion, however, does not exclude the possibility of H-aggregate emission contributing to the red tail of the bulk PLspectrum. We stress that single-chromophore spectroscopy, by itsvery definition, probes only the emission from single chromo-phores; we cannot probe potential H-aggregate emission withoutresorting to well-defined dimer structures (52).

Difference in Polymer Chain Conformation Between Solution and BulkPhase. Why do different chromophores dominate ensemble so-lution and bulk film spectra (Fig. 1A)? The fundamental differ-ence between the two cases is found in the underlying chainconformation, which can be controlled through the polarity ofthe immediate environment of the polymer (30): a “good” sol-vent, or matrix, will lead to optimal chain extension, driving theformation of well-solvated yet disordered spheres or globules. Incontrast, a “poor” solvent promotes collapse of the polymer intotoroidal or rod-like structures (29). Solvent quality for nonpolarorganic compounds like conjugated polymers generally decreaseswith increasing polarity of the solvent. This effect can be repli-cated in the solid state for single chains of P3HT by embeddingthem in different polymer matrices. Fig. 2A compares dilute(10−4 mM, 100 times the concentration used in single-moleculeexperiments) solid solutions of P3HT in a virtually nonpolarZeonex derivative (Zeonex 480) and in poly(methyl-methacrylate)(PMMA), which is more polar. The spectrum of P3HT embeddedin Zeonex (green) closely resembles that of toluene solution,whereas the spectrum of P3HT embedded in PMMA (red)matches the peak and red tail of bulk P3HT. The mismatch inspectra at higher energy likely arises due to the presence of in-completely folded (i.e., blue-emitting) chains in PMMA. Theeffect of solvation is demonstrated in fluorescence micrographsof the films in Fig. 2B. The P3HT/Zeonex film appears uniformin emission, whereas the P3HT/PMMA film shows discrete brightspots. Both images are displayed on the same intensity scale. Inthe P3HT/PMMA film, the background is darker than in P3HT/Zeonex, but the spots are much brighter. The formation of brightspots in PMMA suggests that multiple chains can aggregate to-gether. In the following, we demonstrate that even single isolatedchains in PMMA collapse into ordered structures.Fig. 3 reports measurements of the fluorescence modulation of

single chains at room temperature under rotation of the polari-zation angle, θ, of the exciting laser within the sample plane. Fora straight object, the overall transition dipole should lie along theaxis of the π-orbitals, leading to a strong cosine-squared modu-lation of PL with laser polarization (29). Such a modulation inintensity is sketched in Fig. 3A. The depth of modulation,M ¼ ðImax − IminÞ=ðImax þ IminÞ, provides information on the de-gree of order regarding the transition dipoles of an individualchain. Examples of measurements are shown in Fig. S5. Fig. 3Bcompares histograms for 738 single chains in Zeonex and 587chains in PMMA. Whereas the PL excitation (i.e., absorption) ofthe molecules in Zeonex is mostly weakly polarized due to dis-order, P3HT in PMMA (31) is predominantly ordered with Mpeaking around 0.8. Based on these M values, possible chainconformations are sketched at the top of Fig. 4.

Chromophoric Emission at Room Temperature. Intrachain conforma-tion should have a dramatic impact on single-chain photophysics:

in the random-coil structure, the chromophores will couple onlyweakly to each other, effectively emitting independently. In theordered structures, energy transfer should occur between chro-mophores because interchromophoric distances are reduced.Two distinct experiments in Fig. 4 clearly demonstrate this in-terplay between conformation and photophysics. The Insets inFig. 4A show two representative single-molecule spectra at roomtemperature, measured in Zeonex and PMMA, which closelyresemble the ensemble (Fig. 2A) for the solvated and collapsedstructures, respectively. Fig. 4A displays the PL intensity of asingle chain as a function of time. The fluorescence beam isseparated into two paths by a polarizing beam splitter and re-corded with two separate photodiodes, allowing us to identifyany change in orientation of the emissive transition dipole byquantifying the linear dichroism D as defined in the schematic.An ensemble of different dipole orientations will lead to D ∼ 0,as will a single dipole coincidentally oriented at 45° with respectto both detectors. The example P3HT chain in Zeonex is ap-proximately five times brighter than in PMMA, but exhibitsstrong fluctuations and a gradual overall decrease in intensity(bleaching). The emission intensity does not drop completelyto zero. In contrast, in PMMA, discrete blinking is observed. InZeonex, the linear dichroism fluctuates around zero, exhibitingsmall jumps, which imply the involvement of multiple differentchromophore emitters. In contrast, in the PMMA example, ahigh static D is found: either all dipoles are oriented along thesame axis or only one single chromophore in the polymer isactive at once. The first conclusion is inferred from the exci-tation polarization modulation in Fig. 3B. Below, we demonstrate

500 600 700 800 900

PL

Inte

nsity

Wavelength (nm)

SS

n

P3HT

in toluenein Zeonex

bulk filmin PMMA

10 μm 10 μm

in PMMAin Zeonex

A

B

Fig. 2. Replicating bulk and solution spectra in dilute matrix environmentsat room temperature. (A) The ensemble spectrum of chains dispersed in theinert Zeonex matrix (green) closely matches that of isolated chains in toluene(gray). In contrast, PMMA leads to chain aggregation: the spectrum of P3HTdiluted in PMMA at the same concentration as in Zeonex (red) resembles thebulk film emission (gray). In both cases, the concentration of P3HT is 100times higher than in single-molecule experiments. (B) The effect of matrix-induced phase separation is visible in fluorescence micrographs of the spin-coated films, plotted on the same intensity scale. In Zeonex, the film PLappears uniform, whereas in PMMA bright spots are seen corresponding tothe formation of large multichain aggregates.

Thiessen et al. PNAS Early Edition | 3 of 7

CHEM

ISTR

YPN

ASPL

US

that indeed only one chromophore emits at a time from thesingle chain.The number of multiple chromophores involved in the emis-

sion can be quantified by photon statistics obtained using a cross-correlation between photodiodes detecting the fluorescence di-vided into two pathways with a (nonpolarizing) beam splitter(Fig. 4B) (53–55). Fluorescence is excited by a pulsed laser (488 nm)with a period of 25 ns. If only one photon is emitted by themolecule per laser pulse, it cannot be simultaneously pickedup by both detectors. Therefore, for single emitters, the cross-correlation must drop to zero at zero delay τ between detectorsignals, a phenomenon known as photon antibunching. Becausesuch cross-correlation analysis requires high photon counts, weaverage (53) over 80 and 30 single molecules for PMMA andZeonex, respectively. In Zeonex, the cross-correlation signal atτ = 0 is nearly identical to that at other delays, implying that, onaverage, multiple chromophores on the polymer emit at onceand do not couple efficiently. In PMMA, the cross-correlation atτ = 0 drops to 20% of coincidence photon counts compared withτ ≠ 0, implying predominant single-chromophore emission.Given the large number of chromophores on a chain, this phe-nomenon must result from energy transfer to the lowest-energychromophore and concurrent singlet–singlet annihilation in themultichromophoric assembly (56).

DiscussionCryogenic single-chain spectroscopy of P3HT reveals clear sig-natures of discrete intrachain chromophores. The diversity ofspectral characteristics found in the ensemble when going fromisolated to aggregated chains originates from these chromo-phores. Universal single-chromophore spectra, comparable inshape to those observed in many other materials, scatter overalmost 200 nm, providing a measure of the level of inter-chromophoric inhomogeneous broadening of ∼0.8 eV, greatlyexceeding prior estimates (23). In comparison, the correspond-

ing disorder extracted from identical experiments for LPPP is0.03 and 0.2 eV for PPV (41). Such a level of energetic disorderas found in P3HT is unprecedented in molecular emitters, andthere is no immediately obvious explanation for this scatterbased on conventional models of conjugation in P3HT (57).Ensemble P3HT shows a large spectral shift upon going from

dilute solution to the bulk film (Fig. 1A). Because the transitionbetween solution and film probes such a broad spectral range,intermediate states have been generated by means of solvent-vapor annealing to optimize both absorption and charge trans-port in solar cells (3, 8). Despite this heterogeneity of ensembleP3HT, key elements of the bulk photophysics (22, 23) are con-sistent with the formation of discrete intrachain chromophores.We stress, however, that bulk spectral emission features ex-tending beyond 700 nm may be associated with H-aggregationor charge-transfer phenomena (1), which would only arise in thebulk and cannot be probed readily by single-molecule techni-ques. Notable examples of spectral characteristics of bulk PLthat we cannot probe here are the monotonic changes of spectralshape with temperature (23), and the evolution of the vibronicprogression and temporal red shift in time-resolved PL (58).These features can both be modeled with the H-aggregate pic-ture (23).The same universal spectral features are observed indepen-

dent of energetic separation between excitation and emission.Ultrafast dissipation of excitation energy in P3HT has previouslybeen concluded from photon-echo spectroscopy and was attrib-uted to efficient coupling to vibrations (59). However, the dis-sipation of 0.9 eV, the difference between excitation at 458 nmand emission at 680 nm, would require five quanta of the dom-inant vibration, which seems implausible. Because the emissionintensity of all chromophores is comparable, we conclude thateach chain most likely has very similar absorbing chromophores(presumably short conjugated units), which then populate theemissive chromophore by energy transfer. We note that whenonly one chromophore is present on a polymer chain, such as inβ-phase PF, absorption and emission spectra exhibit the expectednear-perfect mirror symmetry (39). Even though multiple chro-mophores must exist on single P3HT chains, it is reasonable toassume that each single chromophore identified in Fig. 1 by itsemission also has a complementary specific mirror-symmetricabsorbing feature (39). The coexistence of short and long chro-mophores within a single polymer chain, distinguished by theirtransition energy and vibronic coupling strength, has been dem-onstrated previously in polyindenofluorene using single-chainlight-harvesting action spectroscopy (60).Intuitively, one may be inclined to speculate that the most

extended chromophores are also the lowest-energy units. How-ever, slight bending of the backbone, chain torsion, and changesin bond alternation can mask a strict correlation between conju-gation length and lowering of the optical gap (50). Semiempiricalcalculations have suggested that even substantial bending of theπ-system in polythiophene need not necessarily disrupt conjuga-tion (57). Given the high degree of ordering of single-chain P3HTseen in polarization fluorescence modulation (Fig. 3), it istempting to surmise that, in these objects, the π-system is also themost extended (27). However, the fact that the vibronic pro-gression does not depend systematically on transition wavelengthis crucial (Fig. 1D). In PDA, for example, the most extendedchains show a dramatic increase of 0–0 transition oscillatorstrength relative to the 0–1 peak at low temperature, because theextended π-system constitutes an effective linear J-aggregate (43).It is conceivable that exciton self-trapping limits excitonic co-herence and thus the transfer of oscillator strength to the purelyelectronic (0–0) transition. Such an effect has been suggested forβ-phase PF (39), which also shows substantial vibronic coupling innear-perfectly extended chains under cryogenic conditions, incontrast to PDA. It is also conceivable that the broad energetic

0.0 0.2 0.4 0.6 0.8 1.00

50

100

150

Occ

urre

nce

Modulation Depth, M0.0 0.2 0.4 0.6 0.8 1.0

+−

=

θ

Ι

θ

Ι 1 + cos (2θ )

Ι min

Ι minΙ max

Ι max

max

in PMMAin Zeonex

A

B

Fig. 3. Shape dependence of single P3HT chains on matrix material at roomtemperature. (A) The modulation of PL intensity under rotation of the planeof polarization of the exciting laser is recorded. The modulation depth Mprovides a measure of chain extension and order: chains that mix well withthe matrix (i.e., are well solvated) form random-coil structures that absorbany polarization of light. Poorly solvated chains, however, collapse, leadingto ordered anisotropic rod-like structures. (B) Histogram of M values forsingle chains in Zeonex and PMMA. In Zeonex, isotropic structures areformed, whereas PMMA gives rise to highly anisotropic arrangements ofthe chains.

4 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.1307760110 Thiessen et al.

distribution of zero-phonon lines (Fig. 1B) arises from an intrinsicStark shift due to trapped charges (61). However, in this casea correlation should exist between increased red shift and in-creased line width (62), which is not observed; the line widthappears to decrease with decreasing transition energy.The most red-emitting single chromophores (Fig. 1) constitute

the monomolecular precursor states for the red species in bulkP3HT films. Emission in the bulk occurs preferentially fromthese species, populated by energy transfer provided the polymerchain is sufficiently collapsed and ordered, as is the case inPMMA matrices or in the bulk. In this regard, P3HT behaveslike many other conjugated polymers, such as poly[2-methoxy-5-(2′-ethylhexyloxy)-p-phenylene vinylene] (MEH-PPV) (63, 64),with the fundamental difference that interchromophoric ener-getic disorder is at least twice as large. As more and more chainspack together, the quantum yield of fluorescence of the meso-scopic objects within the bulk film is reduced because theprobability of exciton dissociation and charge formation increa-ses with greater average exciton migration distances (24, 33).Although we have identified isolated chromophores that emit atwavelengths of ∼680 nm, a further red shift with respect to thispurely chromophoric transition observed in bulk films may arisedue to the previously described phenomenon of weak H-aggre-gation (22), due to a Stark effect resulting from the electric fieldemanating from local trapped charges (65), or from solid-statesolvatochromic effects (1). Finally, we note that it is both theconformation of the overall chain that changes with polarity(matrix) as well as the energetic distribution of chromophoresthat are responsible for such broad spectral variability. Nano-

scale conformation therefore also affects ensemble absorption,enabling the tuning of absorption spectrum by solvent-vaporannealing (3, 8).The most obvious origin of the energetic spread in universal

single-chromophore spectra lies in structural variations betweensingle chromophores. We conclude that single chromophores areboth flexible and can adopt a wide range of subtly varying con-formations: structural and energetic disorder dominates on thesingle-chromophore level. In contrast, in bulk films, an extraor-dinary high degree of order can be reached (19). We proposethat this unique feature of P3HT arises directly from the disorderon the single-chromophore level: as chains fold back on them-selves or aggregate with other chains, there are always suitablyshaped chromophores present so that a closely packed structurecan be formed. It is conceivable that disorder on the single-chromophore level could actually breed order in the bulk.P3HT constitutes one of the most heterogeneous material

systems explored in the context of organic electronics, which,besides some of the impressive device characteristics (14),accounts for its continuing popularity in the field (13). Never-theless, conventional incremental materials development risksbeing impeded by lack of a robust understanding of primaryphotoexcitations in these systems. Is the primary photophysicsof polythiophenes dominated by intermolecular interactions andH-aggregate species, or does the material actually adhere to theestablished concepts (36, 38, 41) of intrachain chromophores asderived from a wide range of compounds? Our single-moleculeexperiments tend to favor the latter notion, while leaving roomfor features of H-aggregation or other bulk solvation effects in

0 10 20 30 40

-1

0

1

D(t)

Time (s)0 10 20 30 40

0

50

100

150

Pho

ton

Cou

nts

(kH

z)

-50 -25 0 25 500

10k

20k

30k

40k

τ (ns)

Occ

urre

nce

-50 -25 0 25 500

2k

4k

6k

550 750

PL

Int.

λ (nm) 550 750

PL

Int.

λ (nm)

τ

EnergyTransfer

EnergyTransfer

Ain PMMAin Zeonex

in PMMAin ZeonexB

Fig. 4. Excited-state properties of solvated (Zeonex) and collapsed (PMMA) single P3HT chains at room temperature. In solvated chains, the chromophoresemit independently of each other (cartoon). In collapsed chains, energy transfer occurs to the lowest-energy chromophore. (A) Fluorescence intensity andlinear dichroism D, determined by splitting the emission into two orthogonal polarization components as illustrated in the cartoon. In Zeonex, multistepblinking is observed. The emission is only weakly polarized (D ∼ 0), exhibiting slight fluctuations as different chromophores switch on and off. In the PMMAcollapsed-chain configuration, the fluorescence intensity is constant and shows single-step blinking. The emission is strongly polarized (D ≠ 0) with nofluctuations. Representative single-chain PL spectra are shown in the Inset. (B) Photon statistics in emission, given by the cross-correlation of two photo-detectors in the emission pathway. The molecules are excited by laser pulses (25-ns period), controlling photon arrival times τ. In Zeonex, multiple chro-mophores emit, leading to only a 20% dip in photodetector coincidence rates. In PMMA, a dip in coincidence rate by 80% is observed, because no more thanone single photon is emitted for each laser pulse. Based on photon counts and background signal for the two cases, the maximum dip expected for a perfectsingle-photon or two-photon source is indicated by the dashed lines.

Thiessen et al. PNAS Early Edition | 5 of 7

CHEM

ISTR

YPN

ASPL

US

the red tail of the emission spectrum. In contrast to all polymericmaterials studied previously, the energetic heterogeneity ofchromophoric transitions spans almost 1 eV, i.e., much of thevisible spectrum. Such a breadth of possible fundamental tran-sition energies can only be accounted for by an extreme sensi-tivity of electronic structure to conformational variations, whichexplains the dramatic impact processing conditions have on bulkmaterial properties (3, 8).

MethodsP3HT (regioregularity, 95.7%; weight average molecular weight, 65.2 kDa;polydispersity index, 2.2) was purchased from EMD Chemicals and used asreceived. PMMA (weight average molecular weight = 96.7 kDa; number av-erage molecular weight, 47.7 kDa) and Zeonex (Zeonex 480) were obtainedfrom Sigma-Aldrich and Zeon Europe, respectively. Low-temperature single-molecule spectroscopy was carried out in a home-built fluorescence micro-scope as described previously (28). The fluorescence was excited at 458 nmusing a frequency-doubled Ti:sapphire femtosecond laser system (Harmo-niXX, APE; and Chameleon Ultra II, Coherent). To minimize contamination ofthe weak fluorescence signal by background luminescence, single-chromo-phore spectra were recorded from samples deposited, without a polymermatrix, directly on top of SiO2-covered Si wafers. The wafers were mountedin a liquid-helium cold-finger cryostat at 4 K under vacuum. Concentrationseries were performed to ensure that the single-molecule density varied asexpected with solution concentration. Due to the low photon count rates, allexperiments were carried out in spectral imaging mode [i.e., resolving thefluorescence spectrum (28)] rather than under direct 2D imaging. This ap-proach ensured that the observed fluorescence really did originate fromsingle P3HT chains.

Room temperature fluorescence was recorded on a separate microscopesetup based on an Olympus IX71. P3HT chains were embedded in a PMMA orZeonex 480 host matrix according to a previously described procedure (65): (i)Borosilicate glass coverslips were cleaned in a 2% (vol/vol) Hellmanex III(Hellma Analytics) solution, followed by rinsing with MilliQ water. (ii) Theglass coverslips were additionally bleached by a UV-ozone cleaner (Novas-can; PSD Pro Series UV). (iii) The P3HT was diluted in toluene to single-molecule concentration (∼10−12 to 10−13 M) and mixed with a 6% PMMA orZeonex 480/toluene solution. (iv) This solution was dynamically spin-coated

in a nitrogen glove box at 2,000 rpm onto the glass coverslips, leading toa film thickness of ∼200–300 nm with an average P3HT chain density of ∼20individual P3HT chains in a range of 50 × 50 μm2. The sample was in-corporated into a home-built gas flow cell and purged with nitrogen toprevent bleaching by oxygen. Excitation was carried out by a fiber-coupleddiode laser (PicoQuant; LDH-D-C-485) at 485 nm under continuous-waveexcitation for wide-field fluorescence microscopy or under pulsed excitationwith a repetition rate of 40 MHz for confocal fluorescence microscopy andtime-correlated single-photon counting. The excitation light was passedthrough a clean-up filter (AHF Analysentechnik; z485/10), expanded, andfocused (or collimated) via a lens system onto the back focal plane of a 1.35N.A. oil immersion objective (Olympus; UPLSAPO 60XO) through the backport of the microscope and a dichroic mirror (AHF Analysentechnik; z488RDC)for wide-field or confocal excitation. For wide-field microscopy, the fluores-cence signal was imaged on an EMCCD camera (Andor; iXon 897) after anadditional magnification of 1.6× and after passing a fluorescence filter (AHFAnalysentechnik; RS488LP), whereas, for confocal measurements, the fluo-rescence signal was split either by a polarizing beam splitter (Thorlabs; CM1-PBS251) into two orthogonal polarizing detection channels or by a 50/50beam splitter and detected by two avalanche photodiodes (PicoQuant; τ-SPAD-20). The excitation intensities were set to ∼1.5 and 150 W/cm2 for wide-fieldand confocal excitation, respectively. The polarization rotation of the excita-tion beam, confocal detection, and photon statistics analysis were carried outas described previously (65). For the antibunching measurements, we averagedthe fluorescence traces of 80 single chains in PMMA and 30 single chains inZeonex, respectively (53). By considering the count rate, photodetector darkcounts and the background intensity, we estimate the expected magnitudeof the antibunching dip for a perfect single emitter as 10% and 2.5% forP3HT/PMMA and P3HT/Zeonex, respectively. These thresholds are indicatedas dashed lines in Fig. 4B.

ACKNOWLEDGMENTS. We thank D. Lidzey and G. Khalil for helpfuldiscussions. J.M.L. is indebted to The David and Lucile Packard Foundationfor providing a fellowship. A.T. acknowledges the Fonds der ChemischenIndustrie for a Chemiefonds fellowship. D.V.B. acknowledges support froman Energy Frontier Research Center funded by the U.S. Department ofEnergy under Award Number DE-SC0001091. This work was partially fundedby the European Research Council Starting Grant MolMesON (Grant 305020).

1. Pope M, Swenberg CE (1999) Electronic Processes in Organic Crystals and Polymers(Oxford Univ Press, Oxford).

2. You J, et al. (2013) A polymer tandem solar cell with 10.6% power conversionefficiency. Nat Commun 4:1446.

3. Campoy-Quiles M, et al. (2008) Morphology evolution via self-organization andlateral and vertical diffusion in polymer:fullerene solar cell blends. Nat Mater 7(2):158–164.

4. Kim Y, et al. (2006) A strong regioregularity effect in self-organizing conjugatedpolymer films and high-efficiency polythiophene: Fullerene solar cells. Nat Mater 5(3):197–203.

5. Yang XN, et al. (2005) Nanoscale morphology of high-performance polymer solarcells. Nano Lett 5(4):579–583.

6. Barnes MD, Baghar M (2012) Optical probes of chain packing structure and excitondynamics in polythiophene films, composites, and nanostructures. J Polym Sci B50(15):1121–1129.

7. Erb T, et al. (2005) Correlation between structural and optical properties of compositepolymer/fullerene films for organic solar cells. Adv Funct Mater 15(7):1193–1196.

8. Moulé AJ, Meerholz K (2008) Controlling morphology in polymer-fullerene mixtures.Adv Mater 20(2):240–243.

9. Padinger F, Rittberger RS, Sariciftci NS (2003) Effects of postproduction treatment onplastic solar cells. Adv Funct Mater 13(1):85–88.

10. Zen A, et al. (2006) Effect of molecular weight on the structure and crystallinity ofpoly(3-hexylthiophene). Macromolecules 39(6):2162–2171.

11. Salleo A, Kline RJ, DeLongchamp DM, Chabinyc ML (2010) Microstructural characteri-zation and charge transport in thin films of conjugated polymers. Adv Mater 22(34):3812–3838.

12. McCulloch I, et al. (2009) Semiconducting thienothiophene copolymers: design,synthesis, morphology, and performance in thin-film organic transistors. Adv Mater21(10–11):1091–1109.

13. Dang MT, Hirsch L, Wantz G (2011) P3HT:PCBM, best seller in polymer photovoltaicresearch. Adv Mater 23(31):3597–3602.

14. Zhao GJ, He YJ, Li YF (2010) 6.5% efficiency of polymer solar cells based on poly(3-hexylthiophene) and indene-C(60) bisadduct by device optimization. Adv Mater22(39):4355–4358.

15. Roncali J (1992) Conjugated poly(thiophenes)—synthesis, functionalization, andapplications. Chem Rev 92(4):711–738.

16. McCullough RD (1998) The chemistry of conducting polythiophenes. Adv Mater 10(2):93–97.

17. Perepichka IF, Perepichka DF, Meng H, Wudl F (2005) Light-emitting polythiophenes.Adv Mater 17(19):2281–2305.

18. Osterbacka R, An CP, Jiang XM, Vardeny ZV (2000) Two-dimensional electronicexcitations in self-assembled conjugated polymer nanocrystals. Science 287(5454):839–842.

19. Sirringhaus H, et al. (1999) Two-dimensional charge transport in self-organized, high-mobility conjugated polymers. Nature 401(6754):685–688.

20. Niles ET, et al. (2012) J-aggregate behavior in poly-3-hexylthiophene nanofibers.J Phys Chem Lett 3(2):259–263.

21. Yamagata H, Pochas CM, Spano FC (2012) Designing J- and H-aggregates throughwave function overlap engineering: Applications to poly(3-hexylthiophene). J PhysChem B 116(49):14494–14503.

22. Clark J, Silva C, Friend RH, Spano FC (2007) Role of intermolecular coupling in thephotophysics of disordered organic semiconductors: Aggregate emission inregioregular polythiophene. Phys Rev Lett 98(20):206406.

23. Spano FC, Clark J, Silva C, Friend RH (2009) Determining exciton coherence from thephotoluminescence spectral line shape in poly(3-hexylthiophene) thin films. J ChemPhys 130(7):074904.

24. Ruseckas A, et al. (2001) Luminescence quenching by inter-chain aggregates insubstituted polythiophenes. J Photochem Photobiol A 144(1):3–12.

25. Chaudhuri D, et al. (2011) Enhancing long-range exciton guiding in molecularnanowires by H-aggregation lifetime engineering. Nano Lett 11(2):488–492.

26. Cook S, Furube A, Katoh R (2008) Analysis of the excited states of regioregularpolythiophene P3HT. Energy Environ Sci 1(2):294–299.

27. Paquin F, et al. (2011) Charge separation in semicrystalline polymeric semiconductorsby photoexcitation: Is the mechanism intrinsic or extrinsic? Phys Rev Lett 106(19):197401.

28. Lupton JM (2010) Single-molecule spectroscopy for plastic electronics: Materialsanalysis from the bottom-up. Adv Mater 22(15):1689–1721.

29. Hu DH, et al. (2000) Collapse of stiff conjugated polymers with chemical defects intoordered, cylindrical conformations. Nature 405(6790):1030–1033.

30. Huser T, Yan M, Rothberg LJ (2000) Single chain spectroscopy of conformationaldependence of conjugated polymer photophysics. Proc Natl Acad Sci USA 97(21):11187–11191.

31. Adachi T, et al. (2011) Regioregularity and single polythiophene chain conformation.J Phys Chem Lett 2(12):1400–1404.

32. Adachi T, et al. (2012) Conformational effect on energy transfer in single polythiophenechains. J Phys Chem B 116(32):9866–9872.

6 of 7 | www.pnas.org/cgi/doi/10.1073/pnas.1307760110 Thiessen et al.

33. Scheblykin IG, Yartsev A, Pullerits T, Gulbinas V, Sundström V (2007) Excited state andcharge photogeneration dynamics in conjugated polymers. J Phys Chem B 111(23):6303–6321.

34. Sugimoto T, Habuchi S, Ogino K, Vacha M (2009) Conformation-related excitonlocalization and charge-pair formation in polythiophenes: Ensemble and single-molecule study. J Phys Chem B 113(36):12220–12226.

35. Khalil GE, et al. (2011) Spectroscopy and single-molecule emission of a fluorene-terthiophene oligomer. J Phys Chem B 115(42):12028–12035.

36. Bässler H, Schweitzer B (1999) Site-selective fluorescence spectroscopy of conjugatedpolymers and oligomers. Acc Chem Res 32(2):173–182.

37. Rossi G, Chance RR, Silbey R (1989) Conformational disorder in conjugated polymers. JChem Phys 90(12):7594–7601.

38. Shand ML, Chance RR, Lepostollec M, Schott M (1982) Raman photoselection andconjugation-length dispersion in conjugated polymer-solutions. Phys Rev B 25(7):4431–4436.

39. Da Como E, Borys NJ, Strohriegl P, Walter MJ, Lupton JM (2011) Formation ofa defect-free π-electron system in single β-phase polyfluorene chains. J Am Chem Soc133(11):3690–3692.

40. Schindler F, et al. (2005) Counting chromophores in conjugated polymers. AngewChem Int Ed Engl 44(10):1520–1525.

41. Schindler F, Lupton JM, Feldmann J, Scherf U (2004) A universal picture of chromophoresin pi-conjugated polymers derived from single-molecule spectroscopy. Proc Natl Acad SciUSA 101(41):14695–14700.

42. Guillet T, et al. (2001) Emission of a single conjugated polymer chain isolated in itssingle crystal monomer matrix. Phys Rev Lett 87(8):087401.

43. Lecuiller R, et al. (2002) Fluorescence yield and lifetime of isolated polydiacetylenechains: Evidence for a one-dimensional exciton band in a conjugated polymer. PhysRev B 66(12):125205.

44. Feist FA, Basché T (2008) Fluorescence excitation and emission spectroscopy on singleMEH-PPV chains at low temperature. J Phys Chem B 112(32):9700–9708.

45. Mena-Osteritz E, et al. (2000) Two-dimensional crystals of poly(3-alkylthiophene)s:Direct visualization of polymer folds in submolecular resolution. Angew Chem Int Ed39(15):2680–2684.

46. Collins BA, et al. (2012) Polarized X-ray scattering reveals non-crystalline orientationalordering in organic films. Nat Mater 11(6):536–543.

47. Kline RJ, McGehee MD, Toney MF (2006) Highly oriented crystals at the buriedinterface in polythiophene thin-film transistors. Nat Mater 5(3):222–228.

48. Brown PJ, et al. (2003) Effect of interchain interactions on the absorption andemission of poly(3-hexylthiophene). Phys Rev B 67(6):064203.

49. Schott M (2006) The colors of polydiacetylenes: A commentary. J Phys Chem B 110(32):15864–15868.

50. Becker K, et al. (2008) How chromophore shape determines the spectroscopy ofphenylene-vinylenes: Origin of spectral broadening in the absence of aggregation. JPhys Chem B 112(16):4859–4864.

51. Müller JG, Anni M, Scherf U, Lupton JM, Feldmann J (2004) Vibrational fluorescencespectroscopy of single conjugated polymer molecules. Phys Rev B 70(3):035205.

52. Liu S, et al. (2013) Coherent and incoherent interactions between cofacial π-conjugatedoligomer dimers in macrocycle templates. J Phys Chem B 117(16):4197–4203.

53. Hollars CW, Lane SM, Huser T (2003) Controlled non-classical photon emission fromsingle conjugated polymer molecules. Chem Phys Lett 370(3–4):393–398.

54. Lee TH, et al. (2004) Oriented semiconducting polymer nanostructures as on-demandroom-temperature single-photon sources. Appl Phys Lett 85(1):100–102.

55. De Schryver FC, et al. (2005) Energy dissipation in multichromophoric single dendrimers.Acc Chem Res 38(7):514–522.

56. Hofkens J, et al. (2003) Revealing competitive Forster-type resonance energy-transferpathways in single bichromophoric molecules. Proc Natl Acad Sci USA 100(23):13146–13151.

57. Beenken WJD, Pullerits T (2004) Spectroscopic units in conjugated polymers: Aquantum chemically founded concept? J Phys Chem B 108(20):6164–6169.

58. Banerji N, Cowan S, Vauthey E, Heeger AJ (2011) Ultrafast relaxation of the poly(3-hexylthiophene) emission spectrum. J Phys Chem C 115(19):9726–9739.

59. Wells NP, Blank DA (2008) Correlated exciton relaxation in poly(3-hexylthiophene).Phys Rev Lett 100(8):086403.

60. Walter MJ, Lupton JM (2009) Unraveling the inhomogeneously broadened absorptionspectrum of conjugated polymers by single-molecule light-harvesting action spectroscopy.Phys Rev Lett 103(16):167401.

61. Schindler F, Lupton JM, Müller J, Feldmann J, Scherf U (2006) How single conjugatedpolymer molecules respond to electric fields. Nat Mater 5(2):141–146.

62. Müller J, et al. (2005) Monitoring surface charge migration in the spectral dynamics ofsingle CdSe/CdS nanodot/nanorod heterostructures. Phys Rev B 72(20):205339.

63. Bolinger JC, et al. (2012) Conformation and energy transfer in single conjugatedpolymers. Acc Chem Res 45(11):1992–2001.

64. Schwartz BJ (2003) Conjugated polymers as molecular materials: How chain confor-mation and film morphology influence energy transfer and interchain interactions.Annu Rev Phys Chem 54:141–172.

65. Stangl T, et al. (2013) Temporal switching of homo-FRET pathways in single-chromophore dimer models of π-conjugated polymers. J Am Chem Soc 135(1):78–81.

Thiessen et al. PNAS Early Edition | 7 of 7

CHEM

ISTR

YPN

ASPL

US